Cross-section analysis

For the baseline specification, DOMINV and FDI are used as explanatory variables. A linear relationship is assumed resulting in the following regression equation for the cross-section analysis:

i Xi FDIi

DOMINVi

GROWTHi =β1+β2 +β3 +β4 +ε (7)

where i indexes the included countries and X_i represents a set of additional explanatory variables introduced in later specifications

An OLS regression is performed for the cross-section sample of developing economies. Table A.1 in Appendix A lists the 68 countries included in the sample. Since we do not have a theory that can suggest a ‘true’ model of economic growth, it is reasonable to try several different specifications. The test statistics for all slope parameters are calculated by using the White heteroscedasticity consistent variance estimator. The results for the developing economies sample are presented in Table 7.

Table 7 Cross-section analysis, developing economies sample Independent

variable

(1) OLS

(2) OLS

(3) OLS

(4) OLS

(5) OLS

(6) OLS

(7) OLS Constant -3.297

(-4.86)***

-3.206 (-4.77)***

-3.312 (-4.75)***

-2.167 (-2.95)***

-4.705 (-3.79)***

-2.991 (-4.50)***

-1.363 (-0.78) DOMINV 17.051

(4.60)***

16.361 (4.50)***

17.028 (4.62)***

13.051 (3.92)***

18.992 (4.42)***

17.019 (4.60)***

12.454 (2.38)**

FDI 2.942 (3.79)***

2.634 (3.01)***

3.035 (3.41)***

2.157 (2.82)***

2.271 (2.50)**

1.756 (1.94)*

SCHOOL1980 0.801E-01 (0.81)

0.118

(0.86)

GDP1980 0.502E-01

(0.04)

-0.247E-03 (-1.84)*

AFRICA -0.733

(-2.16)**

-0.509

(-0.80)

ASIA 1.803

(4.06)***

2.037

(3.29)***

EFI1980 0.242

(1.15)

-0.892E-01 (-0.52)

WAR -0.270 -0.556

(-0.67) (-1.32) FDI*

SCHOOL

0.620

(3.18)***

R² 0.42 0.44 0.42 0.52 0.47 0.43 0.54

Adjusted R² 0.41 0.40 0.39 0.49 0.43 0.41 0.46

N 68 56 67 68 59 56 52

Note: t-statistics within parenthesis. The symbols *, ** and *** denote statistical significance at the 10, 5 and 1 per cent level, respectively.

The first baseline specification includes the DOMINV and FDI variables. Both variables are significant at the 1 per cent level and have the expected positive signs indicating domestic and foreign investment contributes to host country economic growth.

Specification (2) adds SCHOOL1980 to take into account human capital in the host country.

SCHOOL1980 is insignificant although it has the correct sign. Consequently, there is no indication that the level of human capital in 1980 had a positive effect on economic growth during the period.

The third specification introduces the GDP per capita level in 1980. Since GDP1980 is insignificant, there is no evidence of convergence or divergence in income levels among the developing economies. This result is in line with earlier empirical studies which have mostly failed to find indications of convergence, except for a small group of developed economies.

Specification (4) adds the two regional dummies AFRICA and ASIA. Both variables are significant and have the expected signs, reflecting the exceptionally strong growth performance in East Asia as well as the poor development in sub-Saharan Africa.

The fifth specification introduces the variables EFI1980 and WAR. EFI1980 has the expected positive sign but is insignificant. WAR has the expected negative sign but is also insignificant and consequently there is no indication of WAR having a detrimental effect on host country economic growth. Replacing WAR with an interaction variable between WAR and AFRICA does not result in a significant relationship. It is possible that the identification of which countries have experienced war is too broad and includes several low intensity conflicts unlikely to affect economic growth in the long run.

The sixth specification replaces FDI with the interaction variable FDI*SCHOOL. This variable should be interpreted as representing the joint effect of FDI and the host country absorptive capacity on economic growth. The interaction variable is significant and positive.

This observation reflects an idea frequently referred to in the established theory of innovation.

In specification (7), all of the independent variables are included except FDI*SCHOOL.

FDI, the primary variable of interest, is significant at the 5 per cent level and has a positive effect on economic growth. DOMINV and ASIA are also found to have a significant positive effect on economic growth. The negative and significant coefficient for GDP1980 suggests convergence in income levels between the developing economies. AFRICA is insignificant in this specification but has the expected negative sign.

We also investigate the relationship between FDI and domestic investment. A bivariate OLS regression is performed using DOMINV as the dependent variable and FDI as the independent

variable. FDI is significant at the 1 per cent level and has a positive coefficient, providing a rough indication of complementarity between domestic and foreign investment in the developing economies.

The analysis of the developing economies sample shows that both the FDI and DOMINV variables are robust in several different specifications having a significant positive effect on host country economic growth. The results indicate that both domestic and foreign investment contribute to host country economic growth in developing economies. Whereas the observation for Singapore is an outlier, removal of this observation does not change the results. Appendix D presents the results when the average of the inward stock of FDI per capita (FDIPC) is used as dependent variable. The results are similar.

The paper proceeds to perform a similar cross-section analysis for the sample of developed economies. Analysing the developed economies separately allows us to see whether FDI affects economic growth in developed economies. Table A.2 in Appendix A lists the 22 economies included in the sample. The same baseline specification as for the developing economies (DOMINV and FDI) is used and complemented with GDP1980 in order to check for convergence in income levels. The SCHOOL1980 variable is also used in order to investigate the effect of human capital on economic growth. EFI1980 controls for the quality of host country institutions. The results are presented in Table 8.⁵

Table 8 Cross-section analysis, developed economies sample Independent

variable

(1) OLS

(2) OLS

(3) OLS

(4) OLS

(5) OLS Constant -0.305

(-0.60)

0.597 (0.90)

1.637 (1.38)

1.823 (1.60)

1.818 (1.63) DOMINV 8.421

(3.68)***

5.977 (2.04)*

4.443 (1.45)

3.956 (1.43)

3.841 (1.06) FDI 3.077

(8.74)***

0.438E-01 (0.05)

-0.401 (-0.41)

-0.940 (-0.09)

-0.144 (-0.10)

GDP1980

-0.710E-04 (-1.10)

-0.392E-04 (-0.53)

-0.421E-04 (-0.46) SCHOOL -0.515E-

5Appendix D presents the results when the average inward stock of FDI per capita (FDIPC) is used as the dependent variable. The results are similar.

1980 01 (-1.42)

01 (-1.55)

EFI1980 0.113E-01

(0.10)

R² 0.67 0.20 0.29 0.32 0.32

Adjusted R² 0.63 0.11 0.16 0.15 0.10

N 22 21 21 21 21

Note: t-statistics within parenthesis. The symbols *, ** and *** denote statistical significance at the 10, 5 and 1 per cent level, respectively.

The first specification is the same baseline specification used for the sample of developing economies. FDI and DOMINV are found to have a positive and significant effect on host country economic growth. However, ocular inspection of the observations reveals that Ireland is an extreme outlier. Ireland has attracted very large inflows of FDI; in 2002 the inward stock was close to five times as large as the average stock in the developed economies sample. Ireland’s growth rate is also more than twice as high as the average for the sample. Therefore the first specification is re-run, excluding the observation for Ireland from the sample. Specification (2) reveals that the significant positive effect of FDI on economic growth disappears and DOMINV is also insignificant when Ireland is excluded. Three additional specifications are used but none of these indicate a significant effect on economic growth from FDI. The analysis is not able to find any indications that FDI inflows affect host country economic growth in developed economies.

The cross-section analysis suggests that FDI inflows have a positive effect on economic growth in developing but not in developed economies. However, the small number of observations, particularly for the sample of developed economies, implies that this finding may be unreliable. The results from the cross-section analysis motivate extending the empirical work by using panel data.